Commit | Line | Data |
---|---|---|
96964f3e JB |
1 | #include <stdlib.h> |
2 | #include <stdio.h> | |
3 | #include <string.h> | |
4b580abd | 4 | #include <stdbool.h> |
6a19b8fe | 5 | #include <wchar.h> |
e9a51b68 | 6 | |
6a19b8fe JB |
7 | #define TOKEN_MAX 500 |
8 | ||
9 | struct token_s { | |
10 | const char* type; | |
11 | wint_t value[50]; | |
12 | }; | |
13 | ||
14 | struct token_s token[TOKEN_MAX] = {NULL, 0}; | |
dcfcd9ab JB |
15 | |
16 | FILE *source = NULL, *target = NULL; | |
6a19b8fe JB |
17 | wint_t c; |
18 | unsigned int tokenFound = 0; | |
4b580abd JB |
19 | enum TokenType { |
20 | MOTCLE, | |
21 | SECTION, | |
22 | SSECTION, | |
23 | NPARA, | |
24 | MOT, | |
25 | FIN | |
26 | } tokenType; | |
27 | const char* tokenTypestr[] = { "MOTCLE", "SECTION", "SSECTION", "NPARA", "MOT", "FIN" }; | |
28 | ||
91e46777 | 29 | /* It looks silly to check for each characters but for debugging, it's just the way to go */ |
4b580abd | 30 | bool istAlpha() { |
6a19b8fe JB |
31 | if (c == L'a' || c == L'b' || c == L'c' || c == L'd' || c == L'e' || c == L'f' || c == L'g' || \ |
32 | c == L'h' || c == L'i' || c == L'j' || c == L'k' || c == L'l' || c == L'm' || c == L'n' || \ | |
33 | c == L'o' || c == L'p' || c == L'q' || c == L'r' || c == L's' || c == L't' || c == L'u' || \ | |
34 | c == L'v' || c == L'w' || c == L'x' || c == L'y' || c == L'z' || \ | |
35 | c == L'A' || c == L'B' || c == L'C' || c == L'D' || c == L'E' || c == L'F' || c == L'G' || \ | |
36 | c == L'H' || c == L'I' || c == L'J' || c == L'K' || c == L'L' || c == L'M' || c == L'N' || \ | |
37 | c == L'O' || c == L'P' || c == L'Q' || c == L'R' || c == L'S' || c == L'T' || c == L'U' || \ | |
38 | c == L'V' || c == L'W' || c == L'X' || c == L'Y' || c == L'Z' || \ | |
39 | c == L'.' || c == L'?' || c == L'!' || c == L',' || c == L';' || c == L':' || c == L'-' || \ | |
40 | c == L'\''|| c == L'#' || \ | |
41 | c == L'0' || c == L'1' || c == L'2' || c == L'3' || c == L'4' || c == L'5' || c == L'6' || \ | |
42 | c == L'7' || c == L'8' || c == L'9' || \ | |
43 | c == L'à' || c == L'â' || c == L'ç' || c == L'è' || c == L'é' || c == L'î' || c == L'ô' || \ | |
44 | c == L'ù' || c == L'û' || \ | |
45 | c == L'À' || c == L'Â' || c == L'Ç' || c == L'È' || c == L'É' || c == L'Î' || c == L'Ô' || \ | |
46 | c == L'Ù' || c == L'Û') { | |
4b580abd JB |
47 | return true; |
48 | } | |
49 | return false; | |
50 | } | |
51 | ||
52 | bool isSeparator() { | |
6a19b8fe | 53 | if (c == L'\t' || c == L' ' || c == L'\n') { |
4b580abd JB |
54 | return true; |
55 | } | |
56 | return false; | |
57 | } | |
58 | ||
59 | int scanner() { | |
62426b89 JB |
60 | unsigned int i = 0; |
61 | wchar_t m[6]; | |
4b580abd | 62 | |
4b580abd | 63 | init: |
6a19b8fe JB |
64 | if (c == L' ' || c == L'\t') { |
65 | c = fgetwc(source); | |
4b580abd JB |
66 | goto init; |
67 | } | |
6a19b8fe JB |
68 | if (c == L'\n') { |
69 | c = fgetwc(source); | |
fde9417f JB |
70 | goto initLV1; |
71 | } | |
6a19b8fe JB |
72 | if (c == L'>') { |
73 | c = fgetwc(source); | |
4b580abd JB |
74 | goto MC1; |
75 | } | |
6a19b8fe JB |
76 | if (c == L'=') { |
77 | c = fgetwc(source); | |
4b580abd JB |
78 | goto S1SS1; |
79 | } | |
80 | if (istAlpha()) { | |
6a19b8fe | 81 | token[tokenFound].value[i] = c; |
4b580abd | 82 | i++; |
62426b89 | 83 | c = fgetwc(source); |
4b580abd JB |
84 | goto M1; |
85 | } | |
6a19b8fe | 86 | if (c == WEOF) { |
4b580abd | 87 | goto FIN; |
4b580abd | 88 | } |
62426b89 | 89 | goto error; |
4b580abd JB |
90 | |
91 | MC1: | |
62426b89 JB |
92 | if (c == L'A' && !wcscmp(fgetws(m, 6, source), L"uteur")) { |
93 | wcscpy((wchar_t*)token[tokenFound].value, L">Auteur"); | |
6a19b8fe | 94 | c = fgetwc(source); |
62426b89 | 95 | goto MC2; |
4b580abd | 96 | } |
62426b89 JB |
97 | if (c == L'T' && !wcscmp(fgetws(m, 5, source), L"itre")) { |
98 | wcscpy((wchar_t*)token[tokenFound].value, L">Titre"); | |
6a19b8fe | 99 | c = fgetwc(source); |
4b580abd JB |
100 | goto MC2; |
101 | } | |
62426b89 | 102 | goto error; |
4b580abd JB |
103 | |
104 | S1SS1: | |
6a19b8fe JB |
105 | if (c == L'=') { |
106 | c = fgetwc(source); | |
4b580abd JB |
107 | goto SS2; |
108 | } | |
6a19b8fe | 109 | if (isSeparator() || c == WEOF) { |
d3eb30ef JB |
110 | goto SECTION; |
111 | } | |
62426b89 | 112 | goto error; |
d3eb30ef | 113 | |
4b580abd | 114 | SS2: |
6a19b8fe | 115 | if (isSeparator() || c == WEOF) { |
d3eb30ef | 116 | goto SSECTION; |
4b580abd | 117 | } |
62426b89 | 118 | goto error; |
4b580abd JB |
119 | |
120 | SECTION: | |
121 | tokenType = SECTION; | |
122 | return 1; | |
123 | ||
d3eb30ef JB |
124 | SSECTION: |
125 | tokenType = SSECTION; | |
126 | return 1; | |
127 | ||
4b580abd JB |
128 | M1: |
129 | if (istAlpha()) { | |
6a19b8fe | 130 | token[tokenFound].value[i] = c; |
4b580abd | 131 | i++; |
62426b89 | 132 | c = fgetwc(source); |
4b580abd JB |
133 | goto M1; |
134 | } | |
6a19b8fe | 135 | if (isSeparator() || c == WEOF) { |
4b580abd JB |
136 | goto MOT; |
137 | } | |
62426b89 | 138 | goto error; |
4b580abd JB |
139 | |
140 | initLV1: | |
6a19b8fe JB |
141 | if (c == L' ' || c == L'\t') { |
142 | c = fgetwc(source); | |
4b580abd JB |
143 | goto initLV1; |
144 | } | |
6a19b8fe JB |
145 | if (c == L'\n') { |
146 | c = fgetwc(source); | |
fde9417f JB |
147 | goto initLV1LV2; |
148 | } | |
4b580abd | 149 | if (istAlpha()) { |
6a19b8fe | 150 | token[tokenFound].value[i] = c; |
4b580abd | 151 | i++; |
62426b89 | 152 | c = fgetwc(source); |
4b580abd JB |
153 | goto M1; |
154 | } | |
6a19b8fe JB |
155 | if (c == L'=') { |
156 | c = fgetwc(source); | |
4b580abd JB |
157 | goto S1SS1; |
158 | } | |
6a19b8fe JB |
159 | if (c == L'>') { |
160 | c = fgetwc(source); | |
4b580abd JB |
161 | goto MC1; |
162 | } | |
6a19b8fe | 163 | if (c == WEOF) { |
4b580abd JB |
164 | goto FIN; |
165 | } | |
62426b89 | 166 | goto error; |
4b580abd JB |
167 | |
168 | initLV1LV2: | |
169 | if (isSeparator()) { | |
6a19b8fe | 170 | c = fgetwc(source); |
4b580abd JB |
171 | goto initLV1LV2; |
172 | } | |
fde9417f JB |
173 | if (istAlpha()) { |
174 | goto NPARA; | |
175 | } | |
6a19b8fe JB |
176 | if (c == L'>') { |
177 | c = fgetwc(source); | |
4b580abd JB |
178 | goto MC1; |
179 | } | |
6a19b8fe JB |
180 | if (c == L'=') { |
181 | c = fgetwc(source); | |
4b580abd JB |
182 | goto S1SS1; |
183 | } | |
6a19b8fe | 184 | if (c == WEOF) { |
d3eb30ef JB |
185 | goto FIN; |
186 | } | |
62426b89 | 187 | goto error; |
4b580abd JB |
188 | |
189 | NPARA: | |
190 | tokenType = NPARA; | |
191 | return 1; | |
192 | ||
193 | MOT: | |
194 | tokenType = MOT; | |
195 | return 1; | |
196 | ||
197 | MC2: | |
6a19b8fe | 198 | if (isSeparator() || c == WEOF) { |
4b580abd JB |
199 | goto MOTCLE; |
200 | } | |
62426b89 | 201 | goto error; |
4b580abd JB |
202 | |
203 | MOTCLE: | |
204 | tokenType = MOTCLE; | |
205 | return 1; | |
206 | ||
207 | FIN: | |
208 | tokenType = FIN; | |
209 | return 1; | |
210 | ||
211 | error: | |
212 | tokenType = FIN; | |
213 | return -1; | |
214 | } | |
96964f3e | 215 | |
6a19b8fe | 216 | int main() { |
96964f3e | 217 | // Ouvre le fichier test.txt en lecture seulement (le fichier doit exister) : |
62426b89 | 218 | source = fopen("test.txt", "r+"); |
96964f3e JB |
219 | // Cree et ouvre un fichier target.html en lecture/ecriture |
220 | // avec suppression du contenu au prealable : | |
221 | target = fopen("target.html", "w+"); | |
222 | ||
223 | if (source == NULL) { | |
224 | printf("Impossible d'ouvrir le fichier source\n"); | |
225 | return -1; | |
226 | } | |
227 | ||
228 | if (target == NULL) { | |
229 | printf("Impossible d'ouvrir le fichier target\n"); | |
230 | return -1; | |
231 | } | |
232 | ||
62426b89 | 233 | c = fgetwc(source); // lecture du premier caractere |
4b580abd | 234 | do { |
4b580abd JB |
235 | int scanrt = scanner(); |
236 | if (scanrt == -1) { | |
6a19b8fe | 237 | wprintf(L"Scanner error with token value: %ls\n", token[tokenFound].value); |
4b580abd JB |
238 | exit(EXIT_FAILURE); |
239 | } | |
6a19b8fe | 240 | if (c != WEOF) { |
62426b89 | 241 | wprintf(L"%20s: %ls\n", tokenTypestr[tokenType], token[tokenFound].value); |
4b580abd | 242 | } else { |
62426b89 | 243 | wprintf(L"%20s\n", tokenTypestr[tokenType]); |
4b580abd | 244 | } |
6a19b8fe | 245 | token[tokenFound].type = tokenTypestr[tokenType]; |
28280a4c | 246 | tokenFound++; |
62426b89 | 247 | } while (tokenType != FIN); // tant que la fin du fichier n'est pas atteinte |
96964f3e JB |
248 | |
249 | if (source != NULL) fclose(source); // fermeture du fichier source | |
250 | if (target != NULL) fclose(target); // fermeture du fichier target | |
251 | ||
252 | return 0; | |
253 | } |